* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download The Copines, a Novel Class of C2 Domain-containing, Calcium
Histone acetylation and deacetylation wikipedia , lookup
P-type ATPase wikipedia , lookup
Protein (nutrient) wikipedia , lookup
Model lipid bilayer wikipedia , lookup
Protein phosphorylation wikipedia , lookup
G protein–coupled receptor wikipedia , lookup
SNARE (protein) wikipedia , lookup
Magnesium transporter wikipedia , lookup
Endomembrane system wikipedia , lookup
Protein moonlighting wikipedia , lookup
Signal transduction wikipedia , lookup
Nuclear magnetic resonance spectroscopy of proteins wikipedia , lookup
Homology modeling wikipedia , lookup
List of types of proteins wikipedia , lookup
Protein structure prediction wikipedia , lookup
Intrinsically disordered proteins wikipedia , lookup
Protein mass spectrometry wikipedia , lookup
Western blot wikipedia , lookup
THE JOURNAL OF BIOLOGICAL CHEMISTRY © 1998 by The American Society for Biochemistry and Molecular Biology, Inc. Vol. 273, No. 3, Issue of January 16, pp. 1393–1402, 1998 Printed in U.S.A. The Copines, a Novel Class of C2 Domain-containing, Calciumdependent, Phospholipid-binding Proteins Conserved from Paramecium to Humans* (Received for publication, August 15, 1997, and in revised form, October 7, 1997) Carl E. Creutz‡§¶, Jose L. Tomsig‡, Sandra L. Snyder‡, Marie-Christine Gautier‡§, Feriel Skouri§, Janine Beisson§, and Jean Cohen§ From the ‡Department of Pharmacology, University of Virginia, Charlottesville, Virginia 22908 and §Centre de Genetique Moleculaire, Centre National de la Recherche Scientifique, 91198 Gif-sur-Yvette, France In an attempt to identify proteins that might underlie membrane trafficking processes in ciliates, calcium-dependent, phospholipid-binding proteins were isolated from extracts of Paramecium tetraurelia. The major protein obtained, named copine, had a mass of 55 kDa, bound phosphatidylserine but not phosphatidylcholine at micromolar levels of calcium but not magnesium, and promoted lipid vesicle aggregation. The sequence of a 920-base pair partial cDNA revealed that copine is a novel protein that contains a C2 domain likely to be responsible for its membrane active properties. Paramecium was found to have two closely related copine genes, CPN1 and CPN2. Current sequence data bases indicate the presence of multiple copine homologs in green plants, nematodes, and humans. The full-length sequences reveal that copines consist of two C2 domains at the N terminus followed by a domain similar to the A domain that mediates interactions between integrins and extracellular ligands. A human homolog, copine I, was expressed in bacteria as a fusion protein with glutathione S-transferase. This recombinant protein exhibited calcium-dependent phospholipid binding properties similar to those of Paramecium copine. An antiserum raised against a fragment of human copine I was used to identify chromobindin 17, a secretory vesicle-binding protein, as a copine. This association with secretory vesicles, as well the general ability of copines to bind phospholipid bilayers in a calcium-dependent manner, suggests that these proteins may function in membrane trafficking. Molecular life at the interface of the cell membrane and the cytoplasmic milieu may be regulated by proteins that attach to and detach from the membrane surface in response to signals. Calcium-dependent, membrane-binding proteins may play such a role. Two major protein motifs that regulate calciumdependent interactions with membrane lipids have been extensively characterized: The annexin fold (1, 2), and the C2 domain (3, 4). The annexin fold appears in quartets in the annexin family of proteins. The structure of these proteins, which has been solved (1), consists only of the calcium and lipid binding domains, with the exception of generally short amino-terminal domains that provide additional regulation of the annexin or binding sites for other proteins. In contrast, the C2 domain, for which the structure is also known (3), is a motif that is attached to a diverse array of enzymatic or protein interaction domains to provide calcium and/or lipid regulation of functions inherent in other portions of the protein. Examples of C2 domain-containing proteins include protein kinase C (5), phospholipase C (6), synaptotagmin (7), rabphilin (8), Doc2 (9), and Munc13 (10). The absence of clear enzymatic activities has made the functions of the annexins difficult to determine. Their interaction with membranes can lead to modulation of other membranebinding proteins such as phospholipases (11). They also exhibit a “bivalent” activity in the sense that they can bind to two membranes and therefore draw them together into a complex that is subject to fusion with additional perturbation of membrane structure (2, 12, 13). This activity, as well as relevant localizations of some annexins (14, 15), has led to the proposal that the annexins may mediate membrane-trafficking events. However, some proteins containing C2 domains, such as the cytoplasmic portion of synaptotagmin, are endowed with similar attributes (16), so it is difficult to define activities unique to annexins. We recently attempted to characterize calcium-dependent, membrane-binding proteins from Paramecium tetraurelia because of some of the unique cytological and genetic characteristics of this organism. Our approach depended on the isolation of membrane-binding proteins from EGTA extracts of homogenized cells, an approach that has been very effective in isolating annexins from a wide variety of organisms (17–19). However, the major protein we obtained from Paramecium by this approach was not an annexin but a novel protein with two copies of the C2 domain and one copy of a domain related to the A domain that mediates protein-protein interactions between integrins and their extracellular ligands. Because the Paramecium protein associates with lipid membranes, like a “companion,” we have given the protein a name reflecting this property: copine (pronounced “ko-peen9”), from the French feminine noun copine, which means “friend.” EXPERIMENTAL PROCEDURES * This study was initiated during the tenure of Fogarty Senior International Fellowship FO6 TWO2133 (to C. E. C.) and was supported by National Institutes of Health Grants GM53266, NS31618, and CA40042; by Ministere de l’Education Nationale, de l’Enseignement Superieur, et de la Recherche Grant ACC-SV6 9506004, and by funds from the Centre National de la Recherche Scientifique. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ¶ To whom correspondence should be addressed: Tel.: 804-924-5029; Fax: 804-982-3878; E-mail: [email protected]. This paper is available on line at http://www.jbc.org Purification of Paramecium Copine—P. tetraurelia, wild-type stock d4 –2, was grown at 27 °C in an infusion of wheat grass powder (Pines International, Lawrence, KS) inoculated with Klebsiella pneumoniae and supplemented with b-sitosterol (4 mg/ml) according to Sonneborn (20). Twelve to 18 liters of cell culture at a density of 2,000 – 4,000 cells/ml were harvested by centrifugation yielding a wet cell pellet of 5–10 ml. All subsequent steps were carried out on ice or at 4 °C. The cells were washed once with spring water (Volvic) and resuspended in 3 volumes of homogenization buffer (150 mM NaCl, 50 mM HEPES, pH 7.4, 5 mM EGTA, 50 mM phenylmethylsulfonyl fluoride, and 5 mM leupeptin). The cells were homogenized with 40 strokes of a Potter-Elve- 1393 hjem homogenizer with a tightly fitting pestle. The volume was increased to 50 ml by the addition of homogenizing buffer, and the cells were further homogenized with five strokes of a tightly fitting Dounce homogenizer. By phase microscopy it was found that this procedure resulted in the lysis of all cells, but large organelles appeared intact. The homogenate was centrifuged at 27,000 3 g for 15 min, the pellet was discarded, and the supernatant was centrifuged at 200,000 3 g for 1 h to prepare a postmicrosomal supernatant. Multilamellar liposomes were prepared from 200 mg of a bovine brain lipid fraction enriched to 80% in phosphatidylserine (Sigma product B-1502) by sonication in 150 mM NaCl, 50 mM HEPES-NaOH, pH 7.4, as described previously (21). The vesicles were pelleted at 200,000 3 g for 1 h, and the postmicrosomal supernatant was used to resuspend the lipid pellet with 10 strokes of a tightly fitting Dounce homogenizer. The free calcium level of the suspension was adjusted to approximately 3 mM by the addition of 8 mM CaCl2. The pH, reduced by release of protons from EGTA, was returned to 7.4 by the addition of NaOH. The vesicles were sedimented by centrifugation at 200,000 3 g for 1 h, and the supernatant was discarded. The pellet was resuspended in wash buffer (150 mM NaCl, 50 mM HEPES-NaOH, pH 7.4, 2 mM CaCl2, 50 mM phenylmethylsulfonyl fluoride, 5 mM leupeptin) by 10 strokes with a Dounce homogenizer (first wash). The vesicles were sedimented as above and then resuspended in wash buffer without NaCl (second wash). After sedimentation, the vesicles were resuspended in 10 ml of extracting buffer (25 mM HEPESNaOH, pH 7.4, 10 mM EGTA, 50 mM phenylmethylsulfonyl fluoride) and sedimented as above. The supernatant, containing calcium-dependent lipid-binding proteins, was saved. The lipid pellet was resuspended in extracting buffer again and sedimented as before, providing a second extract. In some cases a third extraction was performed, yielding about 20% additional protein. The extracts from the lipid vesicles were pooled and applied to a Poros Q anion exchange column (PerSeptive Biosystems, Cambridge, MA) equilibrated in 25 mM HEPES, pH 7.4. The flow-through fractions from this column contained the purified copine. Assay of the Lipid Binding and Aggregating Activities of Copine— Phospholipid binding activity of purified copine was determined by incubation of 5 mg of copine with 0.5–1 mg of phospholipid vesicles prepared as for the purification procedure above in the presence of 2 ml of 25 mM HEPES, pH 7.4, 8 mM EGTA, with or without 10 mM CaCl2 or MgCl2 at room temperature. After sedimentation at 100,000 3 g for 30 min, or 12,000 3 g for 10 min, the supernatants were desalted on Sephadex G-25 and lyophilized before applying to SDS-polyacrylamide gels. The pellets were resuspended directly in gel starting buffer and applied to SDS-polyacrylamide gels. The aggregation of lipid vesicles was determined by measurement of the turbidity (absorbance at 540 nm) of the vesicle suspensions in the binding assay after incubation with 5 mg of copine for 30 min to 1 h. Determination of the Partial Sequence of Paramecium Copine—50 mg of purified copine was applied to an SDS-polyacrylamide gel. The 55kDa band was stained with Coomassie Blue and excised, and the protein was isolated by electroelution in an Amicon electroelution cell by application of 100 V for 2 h. SDS was removed from the eluted protein by precipitation of the protein in chloroform and methanol (23). The protein was then subjected to digestion with lysyl endoprotease purified from Achromobacter lyticus (Waco Products, Richmond, VA), and the resulting peptides were isolated by HPLC, as described (24). Peptides were sequenced by Edman degradation on an Applied Biosystems model 470A gas phase sequencer coupled to a 120A phenylthiohydantoin analyzer. Degenerate oligonucleotides were synthesized corresponding to the peptide sequences, favoring codon usage in Paramecium (25). Successful PCR1 reactions (see “Results”) were obtained using the following primers, which incorporated EcoRI restriction sites (underlined). Peptide P55.4 K R V G D D W Primer 59-GCGAATTCAAAAGAGTAGGAGATGATTGG-39 G T T C C C C Peptide P55.2 Coding sequence K E V L T R N 59-AAAGAAGTATTAACTAGAAA-39 G G T G A C C Complementary Primer 59-GCGAATTCTTTCTAGTTAATACTTCTTT-39 T C A C C G G SEQUENCES 1–3 Note that the design of these primers takes advantage of the almost exclusive use by Paramecium of AGA as the codon for arginine (25). Standard PCRs were conducted with Taq polymerase and 30 cycles of 94 °C for 1 min, 52 °C for 2 min, and 72 °C for 3 min. cDNA from P. tetraurelia, prepared as described by Madeddu et al. (26), was used as template. The major product had an apparent size of 900 base pairs on an agarose gel. It was eluted and subcloned into the plasmid Bluescript SK after digestion with EcoRI. Because of the presence of internal EcoRI restriction sites in the PCR product, interpretation of the sequence was not straightforward. Therefore, a second PCR product obtained using the same primers and cDNA sample was subcloned into the TA cloning vector pCRII (Invitrogen, San Diego, CA) for sequencing. The clones obtained by these two procedures arose from different genes (see “Results”). Construction of Expression Vectors for Human Copine I—Two overlapping partial cDNAs corresponding to expressed sequence tags (ESTs) for human copine I were obtained from the American Type Culture Collection (IMAGE Consortium clone identification numbers 51016 and 487481, corresponding to GenBankTM accession numbers .2046 TD[(.2046 TD[(.2046 TD[(E[(.2046 TD)-380(subcloned)ubcAA043485,ed)ubc 4.5 The Copines FIG. 1. Isolation of copine from P. tetraurelia. Fractions obtained during the isolation of copine were examined by SDS-polyacrylamide gel electrophoresis. Lane S, postmicrosomal supernatant. Lane PC, postcalcium supernatant obtained from the postmicrosomal supernatant by adding phosphatidylserine vesicles in the presence of 2 mM free calcium and sedimenting the vesicles. Lane E1, first EGTA extract from the washed lipid vesicles, containing calcium-dependent lipidbinding proteins. Lane E2, second EGTA extract from the washed lipid vesicles, containing additional lipid-binding proteins; C marks the position of copine at 55 kDa. Lane QFT, flow-through of POROS-Q anion exchange column obtained after applying the E1 and E2 extracts. The single band represents purified copine. The numbers on the left are the molecular masses in kDa of the marked migration positions of prestained molecular weight standards. and bound to glutathione-agarose beads. A portion of the fusion protein was eluted from the beads with 150 mM NaCl, 50 mM TRIS-HCl, pH 8.0, 1 mM EGTA, and 10 mM glutathione. The eluate was clarified by centrifugation at 6,000 3 g for 10 min. Less than 5% of the protein bound to the beads initially was obtained in the supernatant, suggesting poor solubility of the fusion protein or poor extraction with glutathione. The supernatant was adjusted to pH 7.3 by the addition of HCl and then incubated in 135-ml aliquots containing 100 ng of protein for 10 min at room temperature in the presence or absence of 3.5 mM CaCl2 and approximately 250 mg of phosphatidylserine-enriched brain lipid vesicles (Sigma B-1502). After centrifugation at 6,000 3 g for 10 min, the pellets and supernatants were examined by Western blotting with the affinity-purified anti-copine I antibodies. General Methods—Standard methods of molecular biology and recombinant DNA technology were as described in Ausubel et al. (31). For Southern blotting (see Fig. 5), hybridization was performed at 55 °C in Church buffer, and washes were at 55 °C in 2 3 SSC, 0.1% SDS, followed by 0.2 3 SSC, 0.2% SDS (31). Macronuclear DNA for Southern blot analysis was prepared as described (32). DNA sequencing was performed using the Sanger method on an ABI Prism 377 automated DNA sequencer. SDS-polyacrylamide gels were run according to Laemmli (22), and Western blotting was performed according to Burnette (33) using horseradish peroxide-coupled secondary antibodies and colorimetric detection with 4-chloronapthol (see Fig. 9) or chemiluminescence (see Fig. 8; Pierce). Protein was assayed by the Bradford (34) method, using bovine serum albumin as a standard. The chromobindin fraction of adrenal medullary cytosol was prepared by affinity chromatography as described (35). RESULTS Isolation of a 55-kDa Phospholipid-binding Protein from Extracts of Paramecium—Calcium-dependent, phospholipid-binding proteins were isolated from the soluble fraction of homogenates of mass cultures of P. tetraurelia by binding to multilamellar vesicles prepared from brain lipid extracts enriched in phosphatidylserine. Fig. 1 shows an SDS gel of fractions obtained from a typical preparation. Lane S is the postmicrosomal supernatant that was prepared in EGTA, representing all of the soluble proteins of the homogenate. Lane PC represents the supernatant remaining after adding lipid vesicles and 3 mM excess free calcium to the postmicrosomal 1395 FIG. 2. Calcium-dependent sedimentation of purified copine. Copine (2.5 mg/ml) was incubated with divalent cations and phosphatidylserine vesicles (250 mg/ml; ;300 mM) in a volume of 2 ml and assayed for sedimentation at 100,000 3 g. SDS-polyacrylamide gels of the resulting supernatants are shown. Lane ENS, copine in 8 mM EGTA without centrifugation (control for total amount of copine in the assay). Lane E, copine in 8 mM EGTA, supernatant after centrifugation. Lane Ca, copine in 2 mM free calcium after centrifugation. The protein has pelleted. Lane E 1 L, copine plus phosphatidylserine vesicles in EGTA after centrifugation. Lane Ca 1 L, copine plus phosphatidylserine vesicles in 2 mM free calcium. Lane Mg 1 L, copine plus phosphatidylserine vesicles in 2 mM free magnesium. C marks the position of the copine band. The migration positions of molecular weight standards are marked on the left with the corresponding masses in kDa. supernatant and then sedimenting the vesicles by centrifugation. Note that this procedure does not cause a visible reduction in any of the bands in the postmicrosomal supernatant. Thus, none of the major proteins of the cytosol appear to bind lipids. The vesicles were then washed twice in high and low ionic strength buffers to remove nonspecifically bound proteins. Lanes E1 and E2 represent the proteins obtained by extracting the brain lipid vesicles with a buffer containing 10 mM EGTA to remove proteins that have bound to the lipids in a calcium-dependent manner. Only small amounts of protein are obtained. In a typical preparation, 70 mg of protein was present in the initial postmicrosomal supernatant, while the first extract from the lipid vesicles contained 360 mg of protein, and the second extract contained 200 mg. The major protein in the EGTA extracts from the lipid vesicles had an apparent mass of 55 kDa. Interestingly, the protein was obtained in greatest amount in the second EGTA extract. This is indicative of a very high sensitivity to calcium, since two washes in 10 mM EGTA were necessary to reduce the concentration of calcium sufficiently (1027 M or less) to remove the protein. The 55-kDa protein was purified to homogeneity (Fig. 1, lane QFT) by passage over the fast protein liquid chromatography anion exchange medium Poros-Q, since it did not adhere to this resin, while other proteins in the extracts were retained. The typical yield of the purified protein was 50 –70 mg, thus representing about 0.1% of the protein in the initial postmicrosomal supernatant. This purified 55-kDa protein is henceforth referred to as “copine,” as discussed in the Introduction. Characterization of the Interaction of Purified Copine with Phospholipids—The small amounts of copine that could be obtained limited the degree of characterization possible. Emphasis was put on comparing the calcium and lipid specificities of copine in relation to those of the annexins and C2 domaincontaining proteins. These specificities were tested in a centrifugation assay using multilamellar brain lipid vesicles. The purified copine was incubated with the vesicles under various conditions. The vesicles were then sedimented, and the supernatants and pellets were analyzed for copine by SDS gel electrophoresis. Fig. 2 illustrates an SDS gel of the supernatants from a TABLE I Binding of Paramecium copine to lipids Table entries indicate whether copine was sedimented (1) or remained in the supernatant (2) when centrifuged at the indicated g force in the presence of EGTA, calcium (Ca), magnesium (Mg), phosphatidylserine vesicles (PS), or phosphatidylcholine vesicles (PC). 100,000 3 g 12,000 3 gseto lipi, P, lipi, P, Ce EGTA typical binding experiment, and Table I summarizes the data from several binding experiments. It was found that calcium alone (i.e., in the absence of phospholipids) caused the copine to pellet at high g force (100,000 3 g), implying that a calciumdependent self-association of the protein was occurring. This self-association was calcium-specific, in that magnesium did not promote the pelleting of copine. At a lower centrifugal force, 12,000 3 g, sufficient to sediment the lipid vesicles, copine was not pelleted unless phospholipids were present, suggesting the copine bound to the vesicles. The copine was evidently not destroyed, since it could be recovered from the pellets, as it was during the initial copine isolation procedure. The association with the lipid vesicles did not occur when magnesium was substituted for calcium (Fig. 2 and Table I) or when phosphatidylcholine vesicles were used instead of phosphatidylserine (Table I). All of these characteristics are typical of annexins or proteins that contain C2 domains. It was also observed that under conditions where copine bound to the lipid vesicles, it exhibited a “bivalent” activity. The protein promoted the aggregation of the vesicles, which could be detected as an increase in the turbidity of the vesicle suspension (Fig. 3). Cloning of a Partial cDNA for Copine—Purified copine was excised from a Coomassie-stained SDS gel, eluted, and subjected to hydrolysis with lysyl-endoprotease to generate peptides for direct sequencing. Six peptides isolated by HPLC were sequenced (Table II). These short sequences did not show significant similarity to known protein sequences. The peptide sequences were used to design degenerate oligonucleotides to amplify corresponding sections of DNA by the polymerase chain reaction from P. tetraurelia cDNA. Since the order of the peptides in the copine sequence was unknown, oligonucleotides in both orientations were prepared corresponding to three peptides (Table I, peptides p55.1, p55.2, and p55.3) and were used in PCR reactions in all possible combinations. Possibly due to the high degeneracy of the oligonucleotides (256 –512-fold), many amplified products were seen on agarose gels of the reaction products. To narrow down the range of candidates for further study, particular attention was paid to amplification products that were of a size corresponding to the sum of the sizes of products using other primer pairs. However, the most promising products obeying such rational rules were determined to be false positives by subcloning and sequencing. Redesign of the oligonucleotides and additional peptide sequence information (p55.4) finally led to the very strong amplification of a 920-base pair product with one set of primers representing portions of peptides P55.2 and P55.4 (Table II; also see “Experimental Procedures”). Subcloning and sequencing of this product verified that it contained a single open reading frame incorporating the sequence of peptide P55.4 used for primer design beyond the region used for the primer per se (Fig. 4). In addition, the PCR product contained the sequence of a third peptide (P55.6, Table II). The primer corresponding to peptide P55.2 was designed using sequence at the N terminus of the peptide and was incorporated at the FIG. 3. Copine promotes calcium-dependent aggregation of phosphatidylserine vesicles. Phosphatidylserine vesicles (250 mg/ ml; ;300 mM) were incubated with 8 mM EGTA or 2 mM free calcium in the presence or absence of 2.5 mg/ml of copine. The A540 of the suspension was measured after 30 min. The initial A540 of all suspensions was approximately 0.15. TABLE II Peptides derived from Paramecium copine The initial lysine in parentheses (K) was assumed to be present due to the use of lysyl endoprotease to generate the peptides. Peptide Sequence P55.1 P55.2 P55.3 P55.4 P55.5 P55.6 (K)NVDGWFGNSDPFLRFYK (K)EVLTRNVVQFVPFAPFSHXLPIIQQE (K)LRDDNRYIFXQXEGK (K)RVGDDWLPVHK (K)KVGDDWLPV (K)KLEFLDQDQK 39-end of the amplified cDNA; thus, no verification of the sequence at this end was possible. However, the amino acid residues encoded by the primer sequence were in the same reading frame as the other two peptides. These data, as well as additional features of the Paramecium and homologous sequences described below, verified that the correct product had been obtained. In retrospect, the most obvious characteristic of the correct PCR product compared with the false positives was that it was obtained in much greater amounts, as indicated by ethidium bromide staining on agarose gels. Comparison of the Paramecium copine sequence with current data bases indicated strong similarity with C2 domaincontaining proteins in the region of the C2 domain. This similarity could be extended if an additional peptide (P55.1 in Table II) were added to the N terminus of the sequence encoded by the partial cDNA (Fig. 4). The presence of the C2 domain provided a rational explanation for the biochemical properties of copine and thus was further evidence that the correct cDNA had been amplified. Two Copine Genes Are Present in Paramecium—Peptides P55.4 and P55.5, in their region of overlap, are identical except for the substitution of one arginine for a lysine (Table II). This suggested that there may be at least two closely related copine gene products. Additional evidence for this came from the cDNA cloning. Two different strategies were used to subclone the copine cDNA product produced by PCR. The two strategies yielded two different copine sequences encoded by different genes The Copines 1397 FIG. 5. Southern blot of Paramecium macronuclear DNA probed with the copine cDNA. DNA from P. tetraurelia was cleaved with the restriction enzymes marked, resolved by electrophoresis in an agarose gel, transferred to nitrocellulose, and probed with the copine cDNA. Two bands of similar intensity are seen in most lanes, suggesting the presence of two closely related copine genes. Numbers on the right indicate the migration distances of mass standards in kilobases (kb). FIG. 4. Sequence of Paramecium copine. The nucleotide sequence of the partial cDNA corresponding to gene CPN1 is numbered 1–920 (GenBankTM accession number U64872). Nucleotide substitutions present in the sequence of the cDNA corresponding to gene CPN2 are indicated above the CPN1 cDNA sequence. The translation of the CPN1 cDNA, using the ciliate genetic code, is given below the nucleotide sequence. Substitutions in the translated sequence of the CPN2 cDNA are given below the amino sequence for CPN1. Portions of the amino acid sequence present in peptides (Table II) are underlined. The peptide at the N terminus has been added to maximize the alignment with other C2 domains (see text). Underlined nucleotide sequences at the 59and 39-ends are in the positions of the degenerate primers used for amplification and are therefore not necessarily identical to the native sequences. (CPN1 and CPN2). In one case the PCR product was cut with EcoRI, taking advantage of the EcoRI sites that had been designed into the oligonucleotide primers to facilitate subcloning. However, due to the presence of internal EcoRI sites in the PCR product and the formation of concatamers during the ligation process a subclone was obtained that contained four pieces of the cDNA in random orientations. To verify that the correct sequence had been visualized from this complex clone and that no small fragments had been omitted, the subcloning was repeated by performing a new PCR and subcloning the uncut insert using the TA cloning method (Invitrogen). The nucleotide sequence of the clone obtained this way (corresponding to the gene we have named CPN1) was 91% identical to the original reconstructed clone (corresponding to gene CPN2). Of the 79 base differences between these clones, only 13 result in changes in the encoded amino acids (Fig. 4), indicating a strong evolutionary pressure to retain the amino acid sequence, which is 96% identical between the two clones. To investigate the total number of closely related copine genes in the Paramecium genome, a Southern blot analysis was performed on Paramecium macronuclear DNA cleaved with various restriction enzymes using the copine cDNA as a probe and conditions of high stringency for hybridization. As seen in Fig. 5, in most digests two genomic restriction fragments bound to the probe with equal intensity, indicating the presence of two closely related genes for copine. The Copine Family of Proteins—Probing current data bases with the copine cDNA sequence revealed the existence of a number of uncharacterized sequences from genomic sequenc- 1398 The Copines FIG. 6. Amino acid sequences of five human copines. The full-length sequence of human copine I is aligned with the partial sequences of human copines II, III, IV, and V. The two C2 domains are shaded in dark gray. A consensus sequence consisting of residues present in at least 50% of 65 previously characterized C2 domains (4) is given in the top line. The A domain is shaded in light gray. Residues that are identical in all sequences (where known) are boxed. X, residues thought to chelate calcium (in the C2 domains) or magnesium (in the A domain). O, conserved histidine present in the copine A domain. The GenBankTM accession number for the copine I sequence is U83246. The accession numbers of representative ESTs corresponding to these copines are as follows: copine I, H19014; copine II, R87434; copine III, N72351; copine IV, H29499; copine V, H09181. ing projects that are similar to that of copine. These included multiple human ESTs, open reading frames found in nematode genomic sequences, and ESTs and genomic sequences from rice and Arabidopsis. Representative cDNAs corresponding to the human ESTs were obtained from the American Type Culture Collection and sequenced. The sequences could be organized into groups representing five different human genes (Fig. 6), which we refer to as human copines I–V. The degrees of identity between the amino acid sequence of copine I and the other human copines, in the known regions of overlap, are as follows: copine II, 60%; copine III, 78%; copine IV, 53%; copine V, 56%. Five different nematode copine genes have also been analyzed (GenBankTM accession numbers Z80223, Z73911, U21317, Z68213, and U28941). An alignment of the inferred amino acid sequences of representative copines from Paramecium, human, nematode, and Arabidopsis is given in Fig. 7. The degree of identity between the human sequence and the other sequences, in the region of overlap, is as follows: nematode, 40%; Arabidopsis, 40%; Paramecium, 33%. The greatest degree of conservation is seen in residues characteristic of the C2 domain and the integrin A domain (see “Discussion”). Cu- The Copines 1399 FIG. 7. Amino acid sequences of copines from divergent organisms. The partial sequence of copine from P. tetraurelia (GenBankTM accession number U64872) is aligned with the full-length human copine I sequence (U83246) and copine sequences from the nematode Caenorhabditis elegans (Z80223) and the green plant Arabidopsis thaliana (AC000106). The two C2 domains are shaded in dark gray. A consensus sequence consisting of residues present in at least 50% of 65 previously characterized C2 domains (4) is given in the top line. The A domain is shaded in light gray. Residues that are identical in the majority of sequences where known (3 of 4, 2 of 3, or 2 of 2) are boxed. X, residues thought to chelate calcium (in the C2 domains) or magnesium (in the A domain). O, conserved histidine present in the copine A domain. To achieve optimal alignment of the Arabidopsis sequence, minor adjustments were made to the intron/exon boundaries previously interpreted in the GenBankTM entry. riously, the conceptualized Arabidopsis sequence begins in the middle of the first C2 domain. No additional exons that could complete the sequence of this domain are apparent in the cosmid clone (GenBankTM accession number AC000106) upstream of the copine gene and downstream of the preceding gene. However, it is not known if this conceptualized protein is indeed expressed. Expression of Recombinant Human Copine I—Portions of two cDNAs corresponding to ESTs for human copine I were ligated into the expression vector pGEX-KG (27) to create vectors for the expression of fusion proteins consisting of GST fused to copine I or domains of copine I (see “Experimental Procedures”). The fusion proteins were expressed to a high level upon induction of the tac promoter in E. coli; however, the 1400 The Copines FIG. 8. Calcium-dependent binding of recombinant human copine I to phosphatidylserine vesicles. Approximately 100 ng of copine I-GST fusion protein was incubated with 250 mg of (;2 mM) phosphatidylserine vesicles (PS) and 1 mM EGTA (EGTA), or 2.5 mM free Ca21 (Ca) as detailed under “Experimental Procedures.” After sedimentation by centrifugation, the supernatants and the lipid vesicle pellets were examined for the presence of copine by Western blotting with an affinity-purified anti-copine I antibody. Migration positions of molecular weight standards are marked on the left with the corresponding masses in kDa. Note that the copine is partially pelleted in the presence of calcium alone and is completely pelleted in the presence of calcium and lipid together. proteins were only poorly soluble (less than 5%). A fusion protein of GST and the C-terminal half of copine I was partially purified by electrophoresis in SDS and used for the immunization of rabbits. The antiserum produced, as well as antibodies purified by adsorption on copine I immobilized on nitrocellulose, reacted strongly with recombinant human copine I and, in a preliminary survey, with a 55-kDa protein in a variety of rat tissues, including heart, lung, kidney, liver, and skeletal muscle. Calcium-dependent Binding of Recombinant Human Copine I to Phosphatidylserine Vesicles—The soluble portion of the full-length copine I-GST fusion protein produced in bacteria was tested for the ability to bind lipid vesicles in a calcium-dependent manner. Similar to the behavior of Paramecium copine, the recombinant human protein bound to and sedimented with vesicles enriched in phosphatidylserine in a calcium-dependent fashion when tested in a centrifugation assay (Fig. 8). Also similar to the Paramecium protein, calcium caused the apparent self-association of the human copine, since a portion of the protein sedimented in a calcium-dependent manner in the absence of lipid (Fig. 8). GST by itself did not sediment with or without lipid in a calcium-dependent manner (not shown), suggesting that the calcium-dependent behavior of the fusion protein was due to the copine moiety. Identification of Chromobindin 17 as Copine—The properties of copine suggested that it might be a member of the chromobindins, a class of soluble proteins that bind to chromaffin granule membranes in the presence of calcium (35). Accordingly, the antiserum to human copine I was used to probe a Western blot of the chromobindin fraction obtained from bovine adrenal medullary cytosol. The band corresponding to the protein catalogued in 1983 as chromobindin 17 (35) reacted strongly with the copine antiserum as well as with antibodies purified from the serum by adsorption on the recombinant GST-copine I A domain fusion protein (Fig. 9). The same result was obtained with antibodies purified by adsorption on the copine I A domain purified after cleavage from the GST (not shown). Since chromobindin 17 has been characterized as a protein that binds to phospholipids as well as intact chromaffin granule membranes (35), is of the appropriate mass (55 kDa), has a particularly high affinity for calcium (35), and, like copine FIG. 9. Identification of chromobindin 17 as a copine. Chromobindins, calcium-dependent chromaffin granule-binding proteins from bovine adrenal medulla, were eluted from a chromaffin granule membrane affinity column with EGTA, electrophoresed in SDS, and transferred to a nitrocellulose strip that was stained in Ponceau S (lane P). 67, the position of annexin VI; CB17, the position of chromobindin 17 (the upper band of a closely spaced doublet); 32–36, the positions of annexins I, II, IV, and V. Chromobindins from a parallel gel were transferred to nitrocellulose and probed with the copine I preimmune (lane PI) or immune (lane I) serum. The single protein reacting with the antiserum (C) corresponds to chromobindin 17 (CB17). A weakly staining band is seen in the preimmune serum but is of higher mobility than the band reacting with the immune serum. Lanes A1 and A2 represent chromobindins stained with antibodies from two different rabbits affinity-purified by adsorption to recombinant copine I. I, has a broad tissue distribution,2 we conclude that chromobindin 17 is a copine. Because of sequence similarities among the various mammalian copines that might result in immunological cross-reaction, we are not certain whether chromobindin 17 corresponds to copine I or another member of the copine family. Chromobindin 17 is frequently present in chromobindin fractions from various tissues as the upper band of a doublet (as seen in Fig. 9). It is possible that the lower band represents another member of the copine family. DISCUSSION Calcium-dependent, membrane-binding proteins have been isolated by affinity techniques from a number of different organisms (e.g. Refs. 17–19 and 35). Typically, these techniques yield a complex mixture of proteins. The proteins obtained in the greatest yield when using lipids or biological membranes as the affinity reagent and the soluble fractions from either plants or animals have always been annexins, although protein kinase C (35), phospholipase C (36), and other unrelated proteins (37) have been obtained in smaller amounts. Paramecium is thus unusual in that the major protein, copine, falls into a different class. Although no annexin has been definitively characterized from Paramecium, there are reports of Paramecium proteins that cross-react with anti-annexin antisera (38).3 It is possible that some of the additional proteins obtained in the lipid-binding fraction in this study are annexins; however, we detected no immunological cross-reaction between these proteins and antisera raised against mammalian annexin I, II, IV, VI, or VII or nematode Nex-1 annexin (data not shown). Copine 2 W. H. Martin, and C. E. Creutz, unpublished observations. D. Kerboeuf, B. Delouche, L.-A. Pradel, and J. Cohen, unpublished observations. 3 The Copines 1401 FIG. 10. Sequence and structural comparison of the copine C-terminal domain and the integrin A domain. Portions of the human copine I sequence (residues 292–512) are aligned with portions of the sequence of the A domain of the a-subunit of human integrin CR3 (residues 8 –191). The alignment was generated using the FASTA program; colons mark amino acid identities; periods marks amino acid similarities. The positions of a-helices and b-strands as determined by crystallography are marked below the A domain sequence. a-Helices and b-strands predicted to be present in copine by the algorithm of Chou and Fasman (42) are marked above the copine sequence. Residues in the A domain sequence that participate in the chelation of magnesium are in boldface type, underlined, and labeled according to residue number. Aligned residues in the copine sequence that may play an analogous role are similarly marked. is the first Paramecium protein to be characterized that possesses C2 domains, thus extending the breadth of this family of lipid-binding proteins to include ciliates. Organization of Domains in the Structure of Copine—Examination of the full-length sequences of a human copine, a nematode copine and a plant copine (Fig. 7) reveals a very interesting domain structure for copine. There are two C2 domains in the N-terminal half of the molecule. Both C2 domains appear to be functional (with the exception of the plant sequence; see “Results”) in the sense that they contain acidic residues implicated in the binding of calcium by the first C2 domain of synaptotagmin (3, 4). In general, there are two distinct topologies for C2 domains (4). Residues corresponding to the first b-strand in the C2 domain of topology type I are found in the same structural position as the eighth b-strand of the C2 domain of topology type II. The topologies can be recognized on the basis of primary sequence depending upon whether the amino acid residues corresponding to this strand are found before (as in synaptotagmin type I) or after (as in phospholipase C type II) the sequences representing the rest of the domain. In copine, both C2 domains are type II. For example, in human copine I these wandering b-strands occur at residues 125–133 and 265–273, at the trailing ends of the C2 domains (Fig. 6). Thus, the organization of the C2 domains of copine obeys two previously recognized generalizations concerning C2 domains (4): 1) if present at the N terminus of a protein, the C2 domain adopts the type II topology; and 2) if two C2 domains are adjacent to one another, they have the same topology. The sequence of the C-terminal half of the copine molecule shows a distant relationship to the A domain found in a number of extracellular proteins or the extracellular portions of membrane proteins such as integrins; von Willebrand factor; complement factor C2; L-type calcium channels; collagens VI, VII, XI, and XIV; and the plasmodial surface protein throm- bospondin-related anonymous protein (39, 40). Interestingly, there are no other examples of A domains in intracellular proteins, so copine represents a unique fusion of a domain typically found in intracellular proteins, the C2 domain, and a domain typically found in extracellular proteins, the A domain. There are no signal sequences or hydrophobic transmembrane sequences evidently coded for by the cDNAs or genomic sequences of the various copine homologs. Furthermore, the Paramecium protein could not be extracted unless the cells were lysed, suggesting that copine is an intracellular protein. The three-dimensional structure of the A domain (also called I domain) from the a-subunit of integrin CR3 (CD11b/CD18) has recently been determined (40). This integrin is a member of the b-2 integrin family and is the major integrin of phagocytic cells. The A domain appears to mediate the binding of the integrin to extracellular ligands in a magnesium-dependent fashion. The similarity between copine and the integrin A domain is strongest in the C-terminal 2⁄3 of the domain (Fig. 10). The hydrophobic motifs at the end of the D and E b-strands of the integrin domain are particularly characteristic in the copine family. In addition, residues that participate in the coordination of magnesium, either directly or through intervening water molecules, in the A domain crystal structure are also present in the copine A domain (Fig. 10). The presence of the corresponding metal-chelating residues suggests that this domain of copine may bind target molecules in a magnesium dependent fashion just as magnesium is required for the binding of the integrin A domain to target proteins in the extracellular matrix. In addition to the sequence similarities, it is notable that the secondary structure predictions obtained for this portion of the copine molecule using the algorithm of Chou and Fasman (41) are in excellent agreement with the actual secondary structure of the corresponding portions of the A domain as determined by crystallography, particularly near the metal-chelating residues 1402 The Copines (Fig. 10). This agreement suggests that this portion of copine may indeed adopt secondary and tertiary structures similar to those of the A domain fold. However, as shown in Fig. 10, the upstream metal-chelating residues (DGSGS in the integrin; hypothetically DFTGS in human copine I) are displaced relative to the downstream chelating residues by an additional 29 amino acids, and the sequence and structural predictions for this intervening region of copine cannot readily be aligned with the A domain. Thus, it is likely that in copine there are significant additions to the loops that may extend from the core of the A domain-like structure. The putative copine A domain has a histidine near the beginning of the domain within a sequence block (SLH) that is conserved in all species, from plants to humans (Fig. 7). No corresponding residue or sequence motif is found in other A domains. It is possible that this histidine is involved in a catalytic function when copine binds a target through its A domain. The a-b-a organization of the structure of the integrin A domain has been recognized as a classical “Rossman fold” (40). This structural motif is also the basis of the formation of nucleotide- or dinucleotide-binding pockets in a large number of intracellular enzymes (42). The copine sequence does not show significant similarity to nucleotide-binding folds that can be recognized by conventional sequence comparison algorithms or by the presence of sequence “fingerprints” characteristic of nucleotide-binding proteins (42). However, it is of great interest that rabbit muscle copine I was recently found to bind to an ATP affinity column in a calcium-dependent manner.4 Thus, if copine does possess an enzymatic function in its A domain, the function may require ATP or a related nucleotide as a cofactor or cosubstrate. Following the putative A domain, the copines have a variable length C-terminal domain that is relatively enriched in prolines. This domain ranges from 34 to 45 residues in the human copines in which it has been sequenced (Fig. 6) and up to 86 residues in one of the nematode copines (Fig. 7). This domain may confer unique characteristics on the different family members and may provide an additional site for protein-protein interactions. Functions of the Copines—The high degree of conservation of the copines, present in plants, animals, and ciliates, suggests they play a fundamental role in eukaryotic cell biology. However, there are no obvious copine homologs encoded by the yeast genome, so they are not necessary for essential processes in yeast. The diversity of copines in any one organism, such as the nematode or the human, suggests that the family members have radiated early in evolution to perform distinct functions. Many proteins that contain C2 domains play roles in membrane trafficking. Well documented examples include synaptotagmin (7), rabphilin (8), and Munc13 (10), all of which have multiple C2 domains (two or three). Additional C2 domaincontaining proteins carry out enzymatic functions in diverse calcium-dependent cell signaling processes, such as phosphorylation by protein kinase C, release of diacylglycerol and inositol trisphosphate by phospholipase C, release of arachidonic acid by cytosolic phospholipase A2 (43), and ubiquitination of proteins targeted for destruction (44). Our identification of copine I as a secretory vesicle-binding protein, as well as the general property of calcium-dependent 4 P. Fadden, M. Campos, and T. A. J. Haystead, unpublished observations. association with membranes, suggests that the copine family of proteins may also be involved in membrane trafficking. Such a role would be consistent with a need for distinct variants of copine to mediate membrane trafficking in distinct pathways. To fully understand the cell biological roles of the copines it will be important to determine if the C-terminal portion of copine acts as an enzyme, perhaps nucleotide-dependent, or, like the integrin A domain, mediates interactions with other proteins. Acknowledgments—We are indebted to Lon Aggerbeck for providing access to fast protein liquid chromatography facilities in his laboratory, to John Shannon for peptide sequence analysis, to Laurence Vayssie for preparation of Paramecium cDNA, and to Jacques Retief for assistance with sequence alignment and presentation. REFERENCES 1. Huber, R., Romisch, J., and Paques, E. (1990) EMBO J. 9, 3867–3874 2. Creutz, C. E. (1992) Science 258, 924 –931 3. Sutton, R. B., Davletov, B. A., Berghuis, A. M., Sudhof, T. C., and Sprang, S. R. (1995) Cell 80, 929 –938 4. Nalefski, E. A., and Falke, J. J. (1996) Protein Sci. 5, 2375–2390 5. Parker, P. J., Coussens, L., Totty, N., Rhee, L., Young, S., Chen, E., Stabel, S., Waterfield, M. D., and Ullrich, A. (1986) Science 233, 853– 859 6. Essen, L.-O., Perisic, O., Katan, M., and Williams, R. L. (1996) Nature 380, 595– 602 7. Sudhof, T. C., and Rizo, J. (1996) Neuron 17, 379 –388 8. Shirataki, H., Kaibuchi, K., Sakoda, T., Kishida, S., Yamaguchi, T., Wada, K., Miyazaki, M., and Takai, Y. (1993) Mol. Cell. Biol. 13, 2061–2068 9. Orita, S., Sasaki, T., Naito, A., Komura, R., Ohtsuka, T., Maeda, M., Suzuki, H., Igarashi, H., and Takai, Y. (1995) Biochem. Biophys. Res. Commun. 206, 439 – 448 10. Brose, N., Hofmann, K., Hata, Y, and Sudhof, T. C. (1995) J. Biol. Chem. 270, 25273–25280 11. Wallner, B. P., Mattaliano, R. J., Hession, C., Cate, R. L., Tizard, R., Sinclair, L. K., Foeller, C., Chow, E. P., Browning, J. L., Ramachandran, K. L., and Pepinsky, R. B. (1986) Nature 320, 77– 81 12. Drust, D. S., and Creutz, C. E. (1988) Nature 331, 88 –91 13. Creutz, C. E. (1981) J. Cell Biol. 91, 247–256 14. Emans, N., Gorvel, J. P., Walter, C., Gerke, V., Kellner, R., Griffiths, G., and Gruenberg, J. (1993) J. Cell Biol. 120, 1357–1369 15. Chasserot-Golaz, S., Vitale, N., Sagot, I., Delouche, B., Dirrig, S., Pradel, L. A., Henry, J. P., Aunis, D., and Bader, M. F. (1996) J. Cell Biol. 133, 1217–1236 16. Damer, C. K., and Creutz, C. E. (1994) J. Biol. Chem. 269, 31115–31123 17. Creutz, C. E., Snyder, S. L., Daigle, S. N., and Redick, J. (1996) J. Cell Biol. 132, 1079 –1092 18. Gerke, V. (1991) J. Biol. Chem. 266, 1697–1700 19. Smallwood, M., Keen, J. N., and Bowles, D. J. (1990) Biochem. J. 270, 157–161 20. Sonneborn, T. M. (1970) Methods Cell Physiol. 4, 241–339 21. Creutz, C. E., Kambouris, N. G., Snyder, S. L., Hamman, H. C., Nelson, M. R., Liu, W., and Rock, P. (1992) J. Cell Sci. 103, 1177–1192 22. Laemmli, U. K. (1970) Nature 227, 680 – 685 23. Wessel, D., and Flugge, D. I. (1984) Anal. Biochem. 138, 141–143 24. Brownawell, A. M., and Creutz, C. E. (1996) Biochemistry 35, 6839 – 6845 25. Martindale, D. W. (1989) J. Protozool. 36, 29 –34 26. Madeddu, L., Gautier, M.-C., Vayssie, L., Houari, A., and Sperling, L. (1995) Mol. Biol. Cell 6, 649 – 659 27. Guan, K., and Dixon, J. E. (1991) Anal. Biochem. 192, 262–267 28. Brownawell, A. M., and Creutz, C. E. (1997) J. Biol. Chem. 272, 22182–22190 29. Olmsted, J. B. (1981) J. Biol. Chem. 256, 11955–11957 30. Frangioni, J. V., and Neel, B. G. (1993) Anal. Biochem. 210, 179 –187 31. Ausubel, F. M., Brent, R., Kingston, R. E., Moore, D. D., Seidman, J. G., Smith, J. A., and Struhl, K. (1989) Current Protocols in Molecular Biology, John Wiley and Sons, Inc., New York 32. Skouri, F., and Cohen, J. (1997) Mol. Biol. Cell 8, 1063–1071 33. Burnette, W. N. (1981) Anal. Biochem. 112, 195–203 34. Bradford, M. M. (1976) Anal. Biochem. 72, 248 –254 35. Creutz, C. E., Dowling, L. G., Sando, J. J., Villar-Palasi, C., Whipple, J. H., and Zaks, W. J. (1983) J. Biol. Chem. 258, 14664 –14674 36. Creutz, C. E., Dowling, L. G., Kyger, E. M., and Franson, R. C. (1985) J. Biol. Chem. 260, 7171–7173 37. Burgoyne, R. D., Cheek, T. R., and Norman, K. M. (1986) Nature 319, 68 –70 38. Knochel, M., Kissmehl, R., Wissman, J. D., Momayezi, M., Hentschel, J., Plattner, H., and Burgoyne, R. D. (1996) Histochem. Cell. Biol. 105, 269 –281 39. Colombatti, A., and Bonaldo, P. (1991) Blood 77, 2305–2315 40. Lee, J.-O., Rieu, P., Arnaout, M. A., and Liddington, R. (1995) Cell 80, 631– 638 41. Branden, C., and Tooze, J. (1991) Introduction to Protein Structure, Garland Publishing, New York 42. Clark, J. D., Lin, L.-L., Kriz, R. W., Ramesha, C. S., Sultzman, L. A., Lin, A. Y., Milona, N., and Knopf, J. L. (1991) Cell 65, 1043–1051 43. Huibregtse, J. M., Scheffner, M., Beaudenon, S., and Howley, P. M. (1995) Proc. Natl. Acad. Sci. U. S. A. 92, 2563–2567 44. Chou, P. Y., and Fasman, G. D. (1974) Biochemistry 13, 222–245